June 16th, 2018

Deep Earth Imaging Future Science Platform currently has a visiting scientist from Stanford University located at our Perth headquarters, Anshuman Pradhan, who recently gave a presentation to the all staff based at the Australian Resources Research Centre (ARRC).


Title: Seismic estimation of reservoir properties with Bayesian Evidential Learning 

Abstract: While seismic elastic property inversion serves as a staple component in reservoir characterization workflows, such an approach has several challenges like rigorous propagation of uncertainties and maintaining computational efficiencies. The Bayesian Evidential Learning (BEL) approach, on the other hand, circumvents these issues by learning a direct stochastic relation between seismic data and reservoir properties and exploiting it to efficiently sample posterior distributions. A theoretical framework is developed for seismic BEL and its efficacy is demonstrated on synthetic examples for estimating low-dimensional sub-seismic reservoir properties and high-dimensional reservoir facies sections from seismic data.


Bio: Anshuman Pradhan is a 4th year Ph.D. candidate in Energy Resources Engineering at Stanford University. He has a B.S. and M.S. degree in Applied Geophysics at Indian Institute of Technology (Indian School of Mines), Dhanbad. His current research interests include Bayesian inverse theory, machine learning, geostatistics, rock physics, seismic imaging and basin modeling.